An independent provider of intelligent driving solutions, Momenta, has commenced trading on the Hong Kong Stock Exchange, marking a significant milestone as the first publicly listed company in the physical AI domain.
The initial public offering was priced at HK$295.6 per share, raising approximately HK$5.89 billion. The stock opened at HK$301, giving the company a market capitalisation of around HK$71 billion.
Specialising in mass-production vehicle solutions, Momenta holds the top position globally among independent providers of city Navigate on Autopilot solutions. Based on projected vehicle sales from March 2025 to February 2026, the company commands a 64.5% market share in this segment, significantly ahead of its nearest competitor. It currently collaborates with 24 global automakers, including nine of the world's top ten. Furthermore, Momenta is expanding its Level 4 autonomous ride-hailing services, rapidly scaling from domestic operations to international markets.
Founded in 2016 by CEO Cao Xudong, a Tsinghua University graduate and former researcher at Microsoft Research Asia, Momenta charted a unique course. A decade ago, when the industry was captivated by the immediate pursuit of Level 4 robotaxis, the company chose a more pragmatic path. Its strategy focused first on developing scalable advanced driver-assistance systems, using vast amounts of real-world driving data to refine its algorithms, thereby building a foundation for the gradual progression toward full autonomy.
In April 2026, Momenta unveiled its R7 large model, a foundational physical AI model built on world models and reinforcement learning. This technology is designed to support the large-scale deployment of Momenta's solutions across passenger vehicles, robotaxis, and autonomous delivery vans.
Since its inception, Momenta has secured multiple funding rounds from a prestigious roster of investors. This includes strategic backers such as SAIC Motor, General Motors, Mercedes-Benz Group AG, Toyota Motor Corporation, NIO Capital, Tencent Holdings Ltd, Alibaba Group Holding Ltd, and Ant Group. Financial investors include Shunwei Capital, Blue Lake Capital, Yunfeng Capital, Granite Asia (formerly GGV Capital's Asia business), IDG Capital, Sinovation Ventures, ZhenFund, and Gaorong Capital. Sovereign wealth fund Temasek Holdings and state-backed entities like E-Town Capital are also among its supporters.
SAIC Motor, the largest external shareholder, holds approximately 8.65% of Momenta post-listing. Based on the opening market cap, this stake is valued at around 5.3 billion yuan, representing a roughly 3.8x return on its cumulative investment of about 1.395 billion yuan.
Other significant shareholders include General Motors (8.58%), Mercedes-Benz Group AG (6.13%), Temasek Holdings (4.51%), Shunwei Capital (4.24%), and NIO Capital (3.6%).
Post-listing, founder Cao Xudong holds a 7.17% stake, while co-founders Xia Yan, Sun Gang, and Sun Huan hold 1.89%, 1.89%, and 1.49%, respectively. Through a dual-class share structure, Cao and several founding team members collectively control approximately 59.85% of the voting rights, ensuring the founding team retains strong influence over the company's long-term strategy and major decisions post-IPO.
The Founder's Path
Cao Xudong, born in 1986 in Gansu province, entered Tsinghua University's engineering mechanics department in 2005. During his undergraduate studies, he developed a keen interest in statistics and extracting knowledge from data.
After earning a direct-entry qualification for a Ph.D. program, Cao made a pivotal decision midway through his doctoral studies. Observing breakthroughs in deep learning for image recognition between 2013 and 2014, he chose to leave the program to pursue a career in artificial intelligence, despite advice from his supervisor.
He subsequently joined the computer vision group at Microsoft Research Asia, often regarded as a training ground for China's AI talent. There, he honed his research skills and learned to translate academic work into practical algorithms, with his contributions applied to products like Xbox and Bing.
A stint at SenseTime as executive R&D director followed, where he led the mass-production deployment of several computer vision algorithms. This experience deepened his understanding of the journey from technology to product and solidified his conviction that the window of opportunity in autonomous driving was opening.
In September 2016, at age 30, Cao co-founded Momenta with several Tsinghua alumni. The name, meaning "initial velocity," signified a fresh start. The early team worked from a windowless office.
At that time, the autonomous driving sector was dominated by the grand narrative of Level 4 robotaxis, with companies like Pony.ai and WeRide attracting significant capital. Cao, however, opted for a path perceived as less glamorous: first developing mass-producible advanced driver-assistance systems, using the data generated to train algorithms, and then progressing toward Level 4 autonomy.
He believed that full autonomy could not be achieved solely through testing with hundreds of vehicles and manual coding. His vision was to provide automakers with limited intelligent driving functions, thereby accumulating vast amounts of real-world driving data—"100 billion kilometres" of it—to fuel the development of full self-driving technology.
This strategy was initially met with scepticism from some investors who viewed the potential of a tier-one supplier as limited compared to the promise of robotaxi services. Nevertheless, Cao remained steadfast. The founding team focused from the ground up on perception algorithms and automated closed-loop systems, relying on their own exploration and validation.
Guided by an intuition for AI's evolution and a belief in data-driven models, the team established the company's core strategy: "One Flywheel, Two Legs." The "flywheel" represents a data-driven, closed-loop for technological evolution, where mass-produced vehicles continuously generate real-world data to refine algorithms. The "two legs" refer to the dual approach of developing revenue-generating, mass-market ADAS solutions to accumulate data, while maintaining a long-term roadmap for Level 4 robotaxis, both leveraging the same underlying algorithm architecture.
This framework took time to materialise. By 2018, despite reaching a valuation exceeding $1 billion and becoming China's first autonomous driving unicorn, Momenta internally resembled a "loose research institute." A headcount expansion to 400 that year consumed significant resources without yielding substantial commercial products.
The period from 2019 to 2020 marked a winter for the autonomous driving industry, with Level 4 commercialisation timelines repeatedly pushed back. While competitors raised funds and advanced aggressively, Momenta receded from public view. Cao made decisive cuts, reducing projects by 60% and relocating the company's operational focus to Suzhou—a move he later considered crucial.
As industry fervour gave way to pragmatism, Momenta's path of prioritising mass production and data closure emerged as a sound strategy. The sales explosion at Tesla's Shanghai Gigafactory in 2020 further convinced automakers of the necessity for advanced intelligent driving.
2021 proved to be a breakout year for Momenta, securing $1 billion in funding—accounting for three-quarters of its total historical fundraising. By 2022, it achieved its first vehicle model launch with a partner. By the end of 2025, Momenta had secured design wins for 170 vehicle models, with 68 already in mass production and over 680,000 units installed. From March 2025 to February 2026, it held a 64.5% market share in China's third-party city NOA market, consistently ranking first.
Major Automakers Place Their Bets
Momenta's funding journey reflects strong backing from the automotive industry's key players. Its early rounds in 2017 included investments from Blue Lake Capital, CDH Investments, SAIC Motor, Sinovation Ventures, and ZhenFund. Shunwei Capital led a subsequent round, with SAIC Motor participating again.
Further 2017 rounds saw investments from NIO Capital, Shunwei Capital, Mercedes-Benz, Temasek, Sinovation Ventures, Granite Asia, and CDH Investments. A series of rounds in 2018 pushed its post-money valuation past $1 billion and introduced heavyweight investors like Tencent Holdings Ltd and China Merchants Capital, making Momenta the highest-valued startup in China's autonomous driving sector at the time.
The pivotal moment for Momenta's capital expansion arrived in 2021. The company completed a series of funding rounds totalling $1 billion, setting a record for the largest single-year fundraising in China's autonomous driving field that year. Alongside follow-on investments from existing backers, new investors included General Motors, Yunfeng Capital, Chery Automobile, Luxshare Precision Industry Co Ltd (via its investment arm), and BYD Company Ltd. This round effectively brought the core global automotive supply chain players into Momenta's shareholder base, providing not just capital but also strategic partnerships with key potential customers.
Unlike many startups reliant solely on financial investors, Momenta's capital structure is distinctly industry-oriented. Nine of the world's top ten automakers are now partners. Among them, seven—including SAIC, GM, Mercedes-Benz, Toyota, BYD, Hyundai, and Chery—are both shareholders and customers.
Subsequent funding rounds from 2022 through 2025 introduced investors such as Alibaba Group, Ant Group, and E-Town Capital. In December 2025, Southeast Asia's largest mobility platform, Grab Holdings Ltd, made a strategic investment.
Gaorong Capital also joined as an investor in a later 2025 round. Partner Xin Wang recalled that the investment decision was made within two weeks of meeting founder Cao Xudong. He highlighted that beyond data and algorithms, Momenta possesses superior batch engineering delivery capabilities—the ability to rapidly replicate and deploy technical solutions across different automakers and vehicle models, ensuring client success. This stems from a mature, reusable engineering and productisation system, a capability born of long-termism that is exceptionally rare.
Tripling Revenue in Three Years and the Ambition for Physical AI
Momenta's revenue has grown exponentially, increasing from 743 million yuan in 2023 to 2.413 billion yuan in 2025, representing a 224.8% increase over three years and a compound annual growth rate exceeding 80%.
This surge is driven by two main factors: rising income from technology development services due to increasing client demand, and a significant increase in royalty fee income as more vehicle models equipped with Momenta's solutions enter mass production.
The company's gross margin has also seen a dramatic improvement, rising from 17.5% in 2023 to 71.6% in 2025. Over the same period, gross profit increased from 130 million yuan to 1.727 billion yuan.
The key driver behind this margin expansion is a qualitative shift in the revenue mix. Momenta's income comprises technology development services and licensing services. Technology development involves project-based, customised work for specific vehicle models before production, which is labour-intensive and carries lower margins. Licensing services involve收取 royalties for each vehicle sold with Momenta's solution post-production.
The proportion of revenue from licensing services jumped from 3.1% in 2023 to 40.1% in 2025. This increase correlates with the number of models reaching mass production, which grew from 8 at the end of 2023 to 26 at the end of 2024, and further to 68 by the end of 2025.
Since the marginal cost of licensing services is extremely low—most costs are incurred during the development phase—the rising share of high-margin licensing revenue has directly lifted the company's overall gross margin.
Comparatively, Momenta's 71.6% gross margin in 2025 leads among listed autonomous driving companies in China. For instance, Horizon Robotics reported a gross margin around 65%, and Black Sesame Technologies around 40% for the same period. Globally, Mobileye Global Inc reported a gross margin of approximately 53% in 2025, putting Momenta 18.6 percentage points ahead.
Despite strong revenue and margin performance, Momenta reported net losses, primarily due to fair value changes of preferred shares and other financial liabilities—a non-cash accounting item unrelated to core operations. Excluding these non-cash impacts, the adjusted net loss narrowed significantly from 1.093 billion yuan in 2023 to 303 million yuan in 2025.
Substantial research and development expenditure is a primary factor in the losses. R&D costs were 1.281 billion yuan, 1.508 billion yuan, and 1.869 billion yuan from 2023 to 2025, accounting for 172.4%, 113.8%, and 77.5% of revenue, respectively. These funds are directed toward core technologies like the physical AI world model, end-to-end autonomous driving algorithms, and full-scene iteration of city NOA.
This heavy R&D investment underpins Momenta's strategic evolution from an "autonomous driving company" to a "builder of foundational physical AI models."
If the boundaries of "digital AI" for pure text and images are becoming clearer, Momenta's next grand ambition lies in enabling AI to understand the rules of motion, causal interactions, and spatial relationships of the real physical world—"physical AI." Autonomous driving, as the current field that successfully combines a data closed-loop with a commercial closed-loop, is arguably the optimal entry point to physical AI.
In April of this year, Momenta officially launched its R7 reinforcement learning world model, achieving a first-in-mass-production milestone. Momenta Vice President Gu Gongyao described R7 as the "ChatGPT moment for physical AI." The core capability of R7 lies in its closed-loop system of massive data pre-training, high-fidelity simulation, and reinforcement learning, enabling the system not just to "see the world" but to "understand" and "predict" it.
From the IPO proceeds, Momenta plans to allocate approximately 60% to R&D, focusing on refining the R7 world model, its data closed-loop toolchain, and AI computing capabilities to solidify its physical AI technological foundation.
On the other "leg" of its strategy, Momenta is accelerating the commercialisation and scaling of its robotaxi services. The company plans to expand its Level 4 fleet to 5,000 vehicles by the end of 2027.
For international expansion, Momenta has established cooperative frameworks with global mobility platforms like Uber Technologies Inc and Grab. It plans to launch commercial robotaxi services in Abu Dhabi and Munich in 2026, with subsequent expansion to cities like Dubai, Singapore, and other locations in Germany. The company has earmarked 20% of the IPO proceeds specifically to support the deployment of its overseas Level 4 robotaxi fleet.
Comments